Reject unclosed DOCTYPE on parsing (#153)

Fix #152

---------

Co-authored-by: Sutou Kouhei <kou@clear-code.com>

lib/rexml/parsers/baseparser.rb

@@ -216,7 +216,12 @@ def pull_event
           x, @closed = @closed, nil
           return [ :end_element, x ]
         end
-        return [ :end_document ] if empty?
+        if empty?
+          if @document_status == :in_doctype
+            raise ParseException.new("Malformed DOCTYPE: unclosed", @source)
+          end
+          return [ :end_document ]
+        end
         return @stack.shift if @stack.size > 0
         #STDERR.puts @source.encoding
         #STDERR.puts "BUFFER = #{@source.buffer.inspect}"
@@ -373,6 +378,9 @@ def pull_event
             @document_status = :after_doctype
             return [ :end_doctype ]
           end
+          if @document_status == :in_doctype
+            raise ParseException.new("Malformed DOCTYPE: invalid declaration", @source)
+          end
         end
         if @document_status == :after_doctype
           @source.match(/\s*/um, true)

lib/rexml/parsers/treeparser.rb

@@ -16,7 +16,6 @@ def add_listener( listener )
 
       def parse
         tag_stack = []
-        in_doctype = false
         entities = nil
         begin
           while true
@@ -39,17 +38,15 @@ def parse
               tag_stack.pop
               @build_context = @build_context.parent
             when :text
-              if not in_doctype
-                if @build_context[-1].instance_of? Text
-                  @build_context[-1] << event[1]
-                else
-                  @build_context.add(
-                    Text.new(event[1], @build_context.whitespace, nil, true)
-                  ) unless (
-                    @build_context.ignore_whitespace_nodes and
-                    event[1].strip.size==0
-                  )
-                end
+              if @build_context[-1].instance_of? Text
+                @build_context[-1] << event[1]
+              else
+                @build_context.add(
+                  Text.new(event[1], @build_context.whitespace, nil, true)
+                ) unless (
+                  @build_context.ignore_whitespace_nodes and
+                  event[1].strip.size==0
+                )
               end
             when :comment
               c = Comment.new( event[1] )
@@ -60,14 +57,12 @@ def parse
             when :processing_instruction
               @build_context.add( Instruction.new( event[1], event[2] ) )
             when :end_doctype
-              in_doctype = false
               entities.each { |k,v| entities[k] = @build_context.entities[k].value }
               @build_context = @build_context.parent
             when :start_doctype
               doctype = DocType.new( event[1..-1], @build_context )
               @build_context = doctype
               entities = {}
-              in_doctype = true
             when :attlistdecl
               n = AttlistDecl.new( event[1..-1] )
               @build_context.add( n )

test/parse/test_document_type_declaration.rb

@@ -53,6 +53,51 @@ def test_no_name
       end
     end
 
+    class TestUnclosed < self
+      def test_no_extra_node
+        exception = assert_raise(REXML::ParseException) do
+          REXML::Document.new("<!DOCTYPE foo [")
+        end
+        assert_equal(<<~DETAIL.chomp, exception.to_s)
+          Malformed DOCTYPE: unclosed
+          Line: 1
+          Position: 15
+          Last 80 unconsumed characters:
+
+        DETAIL
+      end
+
+      def test_start_element
+        exception = assert_raise(REXML::ParseException) do
+          REXML::Document.new(<<~DOCTYPE)
+            <!DOCTYPE foo [ <r>
+          DOCTYPE
+        end
+        assert_equal(<<~DETAIL.chomp, exception.to_s)
+          Malformed DOCTYPE: invalid declaration
+          Line: 1
+          Position: 20
+          Last 80 unconsumed characters:
+          <r>#{' '}
+        DETAIL
+      end
+
+      def test_text
+        exception = assert_raise(REXML::ParseException) do
+          REXML::Document.new(<<~DOCTYPE)
+            <!DOCTYPE foo [ text
+          DOCTYPE
+        end
+        assert_equal(<<~DETAIL.chomp, exception.to_s)
+          Malformed DOCTYPE: invalid declaration
+          Line: 1
+          Position: 21
+          Last 80 unconsumed characters:
+          text#{' '}
+        DETAIL
+      end
+    end
+
     class TestExternalID < self
       class TestSystem < self
         def test_left_bracket_in_system_literal